Tynach

12.3K posts

Tynach

Tynach

@Tynach

Autistic programmer, Linux user, and roleplaying enthusiast. Sometimes interacts with NSFW posts. If you block me, please tell me why so I can do better.

Katılım Aralık 2010
257 Takip Edilen102 Takipçiler
Tynach
Tynach@Tynach·
@keysmashbandit Tried it in incognito mode with adblock and whatnot disabled... And it fscking did the AI thing. Worse, the AI's answer was grossly incomplete and completely skipped over the ways 'word' can be used as a verb. Google needs to stop doing this bullshit. It's making Google worse.
English
0
0
0
4
Tynach
Tynach@Tynach·
@keysmashbandit For me it stopped working with 'define' ages ago.. But I learned that it still works flawlessly with 'meaning'. Works best and most consistently if you manually form the search URL, like this: google.com/search?q=word+… The 'expnd=1' at the end has it show all the meanings at once.
English
1
0
0
13
keysmashbandit
keysmashbandit@keysmashbandit·
"Define "word"" used to give you all definitions of a word. Now it gives you a 5 paragraph small model AI overview for a meaning of the word you didn't even want, and no alternative definitions. That was my most used Google search feature.
keysmashbandit tweet media
English
144
75
2.7K
78.3K
Tynach
Tynach@Tynach·
@Neocatboy Sounds similar to a diffusion language model: arxiv.org/pdf/2502.09992 Bit more of a technical paper, but it helps to know that 'AutoRegressive Modeling (ARM)' refers to 'next token prediction' and that they are replacing that with something that works on multiple tokens at once.
English
0
0
0
4
Neocat
Neocat@Neocatboy·
datacenters halved with this one trick...
Simplifying AI@simplifyinAI

🚨 BREAKING: Tencent has killed the “next-token” paradigm. Tencent and Tsinghua has released CALM (Continuous Autoregressive Language Models), and it completely disrupts the next-token paradigm. LLMs currently waste massive amounts of compute predicting discrete, single tokens through a huge vocabulary softmax layer. It’s slow and scales poorly. CALM bypasses the vocabulary entirely. It uses a high-fidelity autoencoder to compress chunks of text into a single continuous vector with 99.9% reconstruction accuracy. The model now predicts the “next vector” in a continuous space. The numbers are actually insane: - Each generative step now carries 4× the semantic bandwidth. - Training compute is reduced by 44%. - The softmax bottleneck is completely removed. We’re literally watching language models evolve from typing discrete symbols to streaming continuous thoughts. This changes the entire trajectory of AI.

English
1
0
0
17
Tynach
Tynach@Tynach·
@LupoPossum Virgin bucks? Bucks as in, like, male deer?
English
0
0
0
18
Lupo!
Lupo!@LupoPossum·
@Tynach You have to buy it with vbucks I think
English
1
0
1
40
Lupo!
Lupo!@LupoPossum·
That's it I'm getting pregnant
English
10
5
101
1.5K
Tynach
Tynach@Tynach·
@Cointers90 @thatfuckasscat Separate fields of study within psychology are arbitrarily decided. What actually matters is the mechanisms within the brain responsible for behavior, and there's nothing to suggest that these mechanisms are different. That's why both are covered under behavioral psychology.
English
1
0
0
24
(COMMS OPEN!) Cointers el Nuevo Acuñador
(COMMS OPEN!) Cointers el Nuevo Acuñador@Cointers90·
ok, this isn't even a completely false equivalent 😭😭 if you're sexually attracted to anime spiders, you're a fucking weirdo... they have some "human" characteristics but still have the base shape of a fucking spider
English
3
1
34
1.7K
Tynach
Tynach@Tynach·
@Cointers90 @thatfuckasscat That doesn't answer the question. The data is incomplete because it's fundamentally impossible to have complete data about anything. The fact you said it anyway implies you mean something by it, and I want to know what you're trying to say because the literal words you used (c)
English
1
0
0
16
Tynach
Tynach@Tynach·
@nyaraVT Hmmm. Maybe the rise in transphobia is correlated with the rise of LLMs, because transphobia is coming from hallucinating LLMs? (no, I'm not serious.)
English
1
0
1
169
nyara
nyara@nyaraVT·
When someone calls me "he" I feel like they're talking about a person that doesn't exist and I get concerned because they're hallucinating.
English
17
40
673
6.2K
Tynach
Tynach@Tynach·
@rohit_khaire @cmuratori @FelixCLC_ … have played Kohctpyktop more than any other game in my whole life, and have played a lot of Factorio. And it is my somewhat slightly informed opinion that you're full of shit.
English
0
0
0
39
Tynach
Tynach@Tynach·
@rohit_khaire @cmuratori @FelixCLC_ This sounds like the opinion of someone who has never tried to design an integrated circuit before, nor anything else of similar structure and/or scale. I'm not properly educated on the topic, nor have I done it professionally, but I've at least tinkered with Electric VLSI, (c)
English
1
0
0
104
Casey Muratori
Casey Muratori@cmuratori·
While I understand what Jensen was trying to say, if a chip designer said they normally used pencil and paper to do their designs, I would probably assume they were baller AF, not the other way around.
TFTC@TFTC21

Jensen Huang: "If that $500,000 engineer did not consume at least $250,000 worth of tokens, I am going to be deeply alarmed. This is no different than a chip designer who says 'I'm just going to use paper and pencil. I don't think I'm going to need any CAD tools.'"

English
50
35
1.4K
67.6K
Tynach
Tynach@Tynach·
@Cointers90 @thatfuckasscat To know how wide a pencil is, you'd have to know the position of every subatomic particle in the pencil.. And since a single one of those positions would have an infinite number of digits, it would take an infinite amount of data storage space to even store the position of one.
English
1
0
0
20
Tynach
Tynach@Tynach·
@Cointers90 @thatfuckasscat … how accurate IS a ruler even if we assume the middle of each line? Can you measure something so accurately that you know how many carbon atoms across something is, assuming a perfect diamond lattice? Of course not, because it's a fsckin ruler and you're just eyeballing stuff.
English
1
0
0
12
Tynach
Tynach@Tynach·
@Cointers90 @thatfuckasscat … them over. We don't know. And such occurrences are so rare that there is SIGNIFICANTLY more harm and literal evil perpetrated in the name of getting rid of violent video games, than in the name of protecting violent video games.
English
1
0
0
18
Tynach
Tynach@Tynach·
@Cointers90 @thatfuckasscat … became violent was already having other mental problems going on. And it's impossible to tell if they would have become violent anyway, without the game. The game might have been an outlet that delayed their violent outburst, or it might have been the catalyst that tipped (c)
English
1
0
0
12
Tynach
Tynach@Tynach·
@zkurishi The vast majority of modern obscenity laws are actually obscene material, and should be treated as such.
English
0
0
0
17
Tynach
Tynach@Tynach·
@zkurishi Obscenity: media that is formulated to propagate the opinion that a major decrease in the number of nouns in the Universe should occur, where said media is clearly intended to convince people to aid in the real life realization of this decrease in nouns to actually happen.
English
1
0
0
163
くりした善行 (Zenko)🇯🇵無所属/Anti-Censorship/C107 31日ツ01b
くりした善行 (Zenko)🇯🇵無所属/Anti-Censorship/C107 31日ツ01b@zkurishi·
猥褻の定義を恣意的に判断し、取り締れることの理不尽さ。「修正が有れば猥褻では無い」の理屈自体も理解不能だが、現実には修正の仕方まで踏み込んで摘発するかどうかが決められ、そのラインに確たる一貫性がある訳でも無い。 この法律と運用が何を守り続けているのか、過大なものになっていないかは早く国会でも議論されるべき。
STOP!刑法175条@stopcc175

成人向け漫画等で性器に規制が施されています。しかし、この規制（刑法175条が元凶）の科学的な根拠は存在しません。 国会図書館の調査でも、性器修正の必要性を検証した資料は一切見つかりませんでした。 にもかかわらず、修正がないと逮捕される恐れがある現状は、明らかに異常であると言えます。

日本語
32
937
2.5K
133.9K

Keşfet

@keysmashbandit @Neocatboy @LupoPossum @Cointers90 @thatfuckasscat @nyaraVT @rohit_khaire @cmuratori